Applied Sciences in Graphic Communication and Packaging

Inhaltsverzeichnis

1. Image Processing Technology

   1. Frontmatter

   2. A Novel Blind JPEG Image Quality Assessment Based on Blockiness and the Low Frequency Feature in DCT Domain

      Wei Wang, Yuanlin Zheng, Kaiyang Liao, Li Liu, Zhisen Tang

      Abstract

      Blind image quality assessment metrics play an important role in the field of image processing. Blind image quality assessment methods, which are specific to a given type of distortion, are very popular for different image processing applications. JPEG compression is one of the most common image compression methods. In this paper, a support vector regression approach is adopted to assess the quality of JPEG compressed images without reference image. At first, the low frequency feature in DCT domain and the blockiness feature are calculated to present the distortion information of an image. Second, the JPEG dataset is divided into training and testing set randomly. The training set is used to build the SVR model, and the testing set is used to predict the quality score. Finally, combining with MOS or DMOS, the quality score is predicted by SVR model. Extensive experiments on LIVE database demonstrate that the proposed method outperforms the state-of-art methods both on predict accuracy and computational complexity.

   3. The Modeling of 3D Color Topography Based on DEM Data for 3D Printing

      Kaili Zhang, Liuxi He, Xiaochun Wang, Guangxue Chen

      Abstract

      The simulation of the three-dimensional image could not convey the exact three-dimensional (3D) information to people in the practical application; meanwhile 3D terrain model of the physical manufacturing encountered difficulties in mold manufacturing, such as high cost, short life and other bottlenecks. The technology of 3D printing provided a solution for the manufacture of 3D topographic map. Therefore, in this paper, the NSDTF-DEM data was built by colorful 3D model to meet the needs of 3D printing. Based on the Crust algorithm, the regional reconstruction and Delaunay tetrahedral meshing were used to reconstruct the surface of DEM point cloud data. The depth map of the 3D model was obtained by the topography projection of the 3D surface. The color assignment was carried out by using the correspondence between the depth graph gray value and the number of indexed image columns, which was designed to produce the same color texture image as the depth map. This modeling method not only improved the modeling accuracy of the 3D mesh model, but also achieved higher color accuracy.

   4. High Dynamic Range Image Compression Algorithm for Reproduction

      Haiwen Wang, Jie Li, Pengfei Wang, Jiting Wang, Honglin Wang

      Abstract

      High dynamic range image possesses characteristics of wide tone, rich color and detail information, providing better reproduction information. Expound first on principle and features of three high dynamic range image compression algorithms, and the image quality is evaluated by HDR-VDP method. The results indicate that tonal homogenization histogram method can enlarge image tone and contrast, but it causes distortion of color and detail information; Local color adaptive algorithm can maintain better tonal features and detail information, which is compliant for printing output; Algorithm based on iCAM compression can ensure better visual features of human, but the loss of image tones is serious. The comparison of the above three kinds of image compression algorithms provide a better technical approach and method reference for high fidelity reproduction of high dynamic range images.

   5. Research on Image Segmentation of Flexographic Plate Based on OTSU Algorithm

      Yuhui Guo, Wangjian Qiu, Qiang Wang

      Abstract

      This paper uses the improved OTSU algorithm to process the image of dot on flexographic plate and gets the best threshold which can segment the image of dot on flexographic plate accurately. The results prove that the improved OTSU algorithm can obtain the better binarization image of dot, which realizes a high accurate segmentation of the image that the dot area percentage below 10% and deletes the noise in the image. Users calculate the ratio of dot on flexographic plate area by means of the binarization image of dot. The ratio helps people control the quality of dot and error compensation, and makes the prints better.

   6. Feature Extraction and Selection by Machine Learning for Image Quality Assessment

      Yehong Chen, Xiaopeng Sun, Bin Zou

      Abstract

      In this paper, we study on image quality related feature extraction and selection for no reference Image Quality Assessment (IQA). For color images, YIQ space transformation is used for the original RGB image, to carry on the separation of brightness and chromatic information. The 2D log Gabor filters and the Prewitt operators are used to convert these channels into the Phase congruency (PC), the image gradient magnitude (GM), the gradient orientation (GO) and the relative gradient magnitude (RM), which are perception mappings from image as quality features. The final representation of an image is an HVS (human visual system) vector according to the histogram variance of perception mappings. Finally, a mapping from image HVS features to image quality score MOSs is built via an AdaBoost BP neural network on the training data. From the experiments, the quality prediction results in noisier distortion, color images shows that our HVS feature extraction and selection methods are both validated.

   7. Print Inverse Halftoning and Its Quality Assessment Techniques

      Zhangying Jin, Enyin Fang

      Abstract

      The technology of converting halftone image back to the continuous tone image (i.e. inverse halftoning) and method of quality assessment of inverse halftone images are studied. A general method for inverse halftoning of printed image is present in the paper. We use several filters to obtain the smoothing image and edge information of halftone image, and obtain the inverse halftone result by adding the edge information on the smoothing image. The inverse halftone results using the method presented in this paper are visually better than that of using Gaussian low pass filter or that of combining Gaussian low pass filter and median filter. The PSNR of inverse halftone result of amplitude modulation image (value 18.6134) is lower than that of other halftone images, but in fact, the inverse halftone result of amplitude modulation image is visually better. While using the new index of quality assessment D presented in this paper, D value of inverse halftone result of amplitude modulation image is significantly smaller than that of frequency modulation image (the former 5.9506, the latter 0.248), which means the tone feature of inverse halftone result of the amplitude modulation image is more consistent with the continuous image. The inverse halftone results by using the method presented in this paper is good, and the method can be applied to different types of halftone images. And, the index of quality assessment D presented in this paper is helpful to avoiding wrong judgment while PSNR is not consistent with the visual result.

   8. Status and Tendency of Augmented Expression for Paper Map

      Hui Ye, Ruizhi Shi, Shenghui Li

      Abstract

      In the wake of developments in science and technology, a variety of map products comes along. However, there are some inherent defects in the conventional paper map. For instance, it can’t be changed once published; contains limited information; and its real-time performance is also poor. In addition, static rendering mode, lack of interaction and dynamic effect also slow down the development of conventional paper map. The augmented reality makes it possible to present the conventional paper map vividly, and it has wide application prospects. In this paper, the status and progress of augmented expression for paper map are reviewed which may provide reference for future related research.

   9. Unsupervised Image Histogram Peak Detection Based on Gaussian Mixture Model

      Yingping Zheng, Zhijiang Li, Liqin Cao

      Abstract

      Image histogram peak detection is a fundamental technique in digital image processing and relative areas. It has been found that Gaussian mixture model is an effective method to obtain the histogram peaks. However, how to set cluster centers and fit truncation data remain problems that deserve to be explored further. To solve the latter problem, this paper proposes a method consisting of data prediction, unsupervised data fitting and peaks acquisition. Extensive experiments are carried out to demonstrate the performance, and the results prove that our method can improve stability, deal with truncation data, and adaptively find histogram peaks.

   10. Image Retrieval Based on the Multi-index and Combination of Several Features

       Ziwei Tang, Kaiyang Liao, Yuanlin Zheng, Wei Wang, Mengying Liu, Hui Yuan

       Abstract

       Local interest points serve as elementary building blocks in many image retrieval algorithms, and most of them exploit the local volume features using a Bag of Feature (BOF) model. However, the model ignores seriously valuable information about the global features in image and the distribution of the interest points. In this paper, we combine the sift feature and a global color feature. Then, we propose an improved strategy based on the BOF model. Finally, we embed the binary of the sift and color feature in the BOF model. Convincing experimental results on several datasets demonstrate that our proposed method approaches to the state-of-the-art level in image retrieval.

   11. Research on PCB Micrograph Registration Based on FSK Method

       Jie Wang, Qiang Wang, Zheng Gong, Ping Yang

       Abstract

       To realize automatic optical visual inspection (AOI) in the high resolution image for image matching, the extraction and registration of feature points is critical. The traditional image registration algorithm is time-consuming, computationally intensive and ineffective. In this paper, an optimization algorithm based on FAST, SURF and K-Means algorithms is proposed to optimize the feature point selection and reduce the matching point pairs. The algorithm uses the FAST algorithm to extract the feature points, and then generates the SURF descriptors according to the information in the fields around the feature points, and the K-Means algorithm is used to cluster the feature points. Finally, the error matching points are eliminated by the consistency check. The experimental results show that the proposed algorithm has high speed and precision in extracting the feature points.

   12. Study on Digital Acquisition Method of High Dynamic Range Image

       Weiyan Zhang, Qiang Wang, Ping Yang, Jing Cao

       Abstract

       Based on the optimal image exposure set, the dynamic range of the image acquired by the digital camera is extended. In this paper, a high dynamic range image acquisition method is proposed to control the linear dynamic range of single segment exposure by gray scale. When letting the linear dynamic range of the gray scale’s correct single exposure as a benchmark, the range of image is determined. However, the details of the image are seriously missing. To solve this problem, the whole gray scale samples are segmented with the linear part being exposed firstly, leaving high key and shadow for next exposure. The experimental results show that this method can effectively extend the dynamic range of the acquired image, while eliminating the huge image data set by the premise of acquiring the static scene, and increase the detail information of the image’s shadow and high key.

   13. Online Image Processing Based on Web

       Zhongyue Da, Wenjie Yang, Shuai Shao, Xinggen Qian, Shun Pan, Yan Li, Bingjie Shi

       Abstract

       In this paper, online image processing method is researched and online image processing based on server designed. First, system requirements analysis on basis of background is introduced. Then, java language, AJAX (Asynchronous JavaScript and XML) and digital image processing technology are used, and Tomcat is used as a server, Hibernate + Spring + Dorado is used as a system architecture, JavaScript and Dorado view control is used as display technology. Finally, the following three functions are implemented: (1) image upload, which is achieved by using the upload file API; (2) online image browsing, which achieves the client browser server images; (3) online image processing, including the saturation, contrast, brightness, cutting and other functions. The test results show that the system can meet the basic image processing and require very few client resources.

   14. High Image Quality Tone Adjustment Methods for Binary Images

       Jing Geng, Congjun Cao, Weiguo Wu

       Abstract

       Graduation adjustment of a binary image is different from the grayscale image. Printing density measured value of binary image dots is regarded as the gray value of a grayscale image. In this study, based on the boundary pixel conversion around a dot, we propose the graduation adjustment scheme for image shading information. The scheme shows the effect of suppressing image blurring and folding noise.

   15. The Effect of Surface Structured Light’s Phase-Shift on 3D Measurement Accuracy

       Shun Pan, Dan Zhu, Yonghao Xiao, Jiong Liang, Yusheng Lian, Kun Hu

       Abstract

       In the field of 3D vision measurement, surface structured light measurement is a new active technology, which can be widely used in the field of non-contact 3D modeling. Through the analysis of the mathematical model of the surface structured light system, a key point for the accuracy of 3D measurement is whether the spatial coordinates of pixels in encoding patterns can be accurately obtained, which is also the standard to judge the quality of coding method. Therefore, the paper adopts a coding method combining Gray code with Phase-Shift patterns. Simultaneously, a 3D vision measurement system composing a single digital camera and a single digital projector is constructed. This method is proved not only to improve the accuracy of 3D measurement, but also to be more efficient than only Gray coding method, which is apt in oil painting scanning with 3D reproduction, incomplete cultural relic’s repair and construction of 3D geological model etc.